EOSA, VINAL AND McDANIELI THE SILVER VOLTAMETER 43 



the filtrate will show a neutral reaction toward the various indi- 

 cators if the original AgNOs contained no uncombined acid or 

 base. By using iod-eosine as an indicator so little as 1 part in 

 1,000,000 of nitric acid (or alkah) added to the original AgNOs 

 can be detected in the filtrate by titration with y/oo- HNO3 or 

 NaOH. The results have been expressed in terms of parts per 

 million of the nitric acid. 



For the detection of the more resistant forms of reducing 

 impurities (principally cellulose hydrates from filter paper) and 

 colloidal silver a titration test with y^o KMn04 solution has 

 proved very valuable. The slightly acidified crystals of silver 

 nitrate are fused as will be described for purifying the salt and 

 dissolved in water to form a 66 per cent solution. Ten cubic 

 centimeters of this solution is acidified by adding 1 cc. of con- 

 centrated nitric acid (free from nitrous acid) and the -fwoi) potas- 

 sium permanganate solution added in 0.5 cc. portions until the 

 pink color persists for five minutes or more. The number of 

 cubic centimeters thus required the authors have called the ' ' per- 

 manganate number" of the silver nitrate. In test cases this has 

 been found to be proportional to the amounts of oxycelluloses 

 previously added to the silver nitrate and very minute traces 

 can thus readily be detected. It is not claimed that the organic 

 material is necessaril}^ oxidized completely but nevertheless the 

 method is much more sensitive than any method of direct com- 

 bustion would prove under the circumstances as well as being 

 more expeditious. By titrating these concentrated solutions of 

 silver nitrate less than 0.001 per cent of colloidal silver may be 

 detected, a result that is of particular value in showing when 

 over fusion of the salt has occurred in its preparation. 



The water used was twice distilled in block tin and kept in 

 bottles of especially prepared resistant glass. Samples of it were 

 frequently tested for its conductivity and alkalinity. These were, 

 on the average, about 1.3 X 10~'' reciprocal ohms and 0.6 parts 

 in 1,000,000 respectively. The water in the stills was boiled by 

 steam under pressure thereby avoiding the possibility of con- 

 tamination by COo which may greatly increase the deposit in the 

 voltameter. 



